Monday, March 09, 2015

An Impending Shakeout In Library Prep?

My ABGT teleconference-based pieces all had a theme of library preparation.  Library prep has never been as flashy as instrument performance, but is clearly critical.  A library-free sequencing technology remains a distant dream, so DNA (or RNA) must go through a series of preparative steps prior to being loaded on the sequencer.  The dominant library prep molecular biology for clonal sequencing systems consists of shearing the DNA mechanically, making flush ends with a repair mix, adding 3' runs of A and then ligating primers and finally using PCR to amplify the material.  
Mechanical shearing can be replaced with enzymatic shearing (or perhaps even chemical, though I'm unaware of chemical shearing being used in production).  For RNA of different sorts,
some upstream steps are added to convert the RNA to DNA, perhaps with a depletion at some stage of hyperabundant species such as rRNA.  This conversion may, with different levels of success, mark which strand was sense and which antisense. The transposase-based Nextera protocol represents the most drastic departure from this paradigm, enzymatically eliminating all the steps prior to PCR. 

Because the molecular biology is so widely understood, the library preparation space has few barriers to entry. Consequently, a large number of vendors have launched kits for Illumina preparation, with smaller numbers supporting Ion Torrent and (back in the days of yore) the Roche 454 platform.These vendors constantly jockey to differentiate themselves on lower price, fewer steps, faster workflows, friendliness to automation on common liquid handlers, smaller required input amounts, higher accuracy in marking sense RNA, more barcodes, etc. Given the size of the booths that some of these vendors have at meetings such as AGBT, one can have quite a successful business despite the jostle.  

For we in the genomics community, this has translated to steady improvement in what samples can be sequenced and the number which one can afford to sequence.  Grumble as I might that the cost per library remains too high (and it remains too high!), particularly for large microbial projects, or that the cost reduction over time is on the bunny slope compared to the double diamond cost curve of raw sequencing, library preparation has continued to improve.  

As users we'll always cheer improvements in library prep, such as several highlighted at AGBT, but for vendors in the space those spotlights look an awful lot like oncoming headlights.

If you are a library prep vendor, the buzz around 10X Genomics should be modest cause for concern; if you spot a GemCode at one of your customers it could be dismissed as a sign they are trying to stay on the leading edge. Since libraries will cost $500 a pop, well above many other kits, you'd probably find labs still willing to listen to a pitch for $100 (or lower preps).  For labs dealing with minuscule inputs, such as clinical labs, GemCode may be attractive with only 1 ng input, but presumably the other kits for such small amounts (which may include amplification) can sell at a lower price point.

However, seeing a NeoPrep on a lab visit is a cause for alarm.  Illumina is launching kits with very good specs (25 nanogram input), but more importantly the instrument makes everything simple. Illumina's long game is certainly to target the clinical market, which will require such simple automation so every reference lab need not hire an experienced molecular biology jockey, but in the 
short term Illumina is planning to market NeoPrep strongly to the research market.  I expect it will find a ready market for core labs and other groups running modest levels of samples; the simplicity factor will be a big win.  Illumina is also pricing the kits aggressively, so they will be hard to undersell.  But worst of all, if you're anyone else you the competing vendor are utterly locked out: Illumina controls the proprietary cards for the device, creating an insurmountable barrier. Illumina's primary goal with NeoPrep would seem to be to support the budding clinical market, but if it corners the research market along they way they won't mind.

Now, some vendors are poised to compete.  To give just one example, Kapa's single tube HyperPlus kit offers a lot of attractive aspects especially simplicity and automation friendliness. Big labs processing thousands of samples a year may prefer 96 well plate automation solutions over the NeoPrep's 16-at-a-time approach. So there will probably still be some big fish which can be landed, but they will be hotly pursued by every vendor, including Illumina.

The next wave of sequencers is unlikely to offer many new opportunities.  Should QIAGEN ever actually launch their box, they are unlikely to make it easy for anyone else to play; after all, QIAGEN is historically a reagent and kit company.  BGI is making noises about some killer sequencer, probably based on Complete Genomics' technology, but so far there is little more than rumor and broad performance claims to go on. PacBio preps haven't been a big enough market to tempt anyone else; even with growth in interest in that platform.  Worst of all is Oxford Nanopore, which requires multiple proprietary reagents, completely locking out any competition. 

So there is a very real possibility that library vendors will be chasing fewer and fewer opportunities.  Small volume labs that can't/won't spring for a NeoPrep could in aggregate represent a sizable market, but that's hard work.  Large volume labs will huge wins for a lucky few.  There will always be markets for the bulk reagents, as groups experiment with unusual library preps.  But selling individual reagents won't command the same markup as a finished kit.  In general, I would expect there would be pressure for the smaller shops to be acquired by bigger reagent houses which can drive economies of scale.  I don't wish ill of any of the vendors, but don't be surprised to see some of them fading away.

No comments: